Crowd-sourcing the British Bronze Age: Initial experiences and results from the MicroPasts Project Neil Wilkin, Andrew Bevan, Chiara Bonacchi, Adi Keinan-Schoonbaert, Daniel Pett & Jennifer Wexler Introduction Neal Ascherson (2002) has argued that some nations are ‘tidy with their past’, while others leave theirs ‘unsorted’ for ‘scavengers [to] wander, pulling up interesting fragments’ (Ibid., vii). Ascherson reassures us that the latter attitude is nothing to be ashamed of, given that the lack of a ‘commanding ‘story’ which dictates how the past should be understood allows space for imagination and originality’ (Ibid., viii). But there is arguably a productive tension to be found in Ascherson’s duality: a means of disrupting traditional approaches by constructing original ‘stories’ that are both grounded in data and open to a democracy of voices; enabling small, individual discoveries and large-scale data-led analyses/interpretations (typically by professional archaeologists). This requires databases that are both ordered and open. Online datasets of this type are increasingly popular through resources such as the UK Archaeology Data Service (ADS)1, the UK Data archive2, Open Context3, and The Digital Archaeological Record (tDAR)4. The importance, potential and conceptual challenges of making these datasets available digitally is the topic of considerable, productive discussion (e.g. Onsrud and Campbell 2007; Newman 2011; Bevan 2012; Hole 2012; Cooper & Green forthcoming). The MicroPasts project, an AHRC-funded collaboration between the Institute of Archaeology at University College London (UCL) and the British Museum (BM), provides a multi-faceted web platform that brings together full-time academic researchers, museum staff, volunteers and a range of interested parties in order to collaborate on projects that have previously proved difficult to publicise, develop and fund despite their potential worth and perhaps because of the difficult or laborious nature of the task at hand. Summaries of the project, its aims, objectives and initial outcomes have been published elsewhere (Bevan et al. 2014; Bonacchi et al. 2014). An initial aim of the project was to work towards making a long-standing dataset, the National Bronze Age Implement Index (or NBAI, also known as NBII), available to a wider public audience and to engage the public in the course of asking for their help in the process. This contribution outlines the particular relevance of the project to the period-specific demands of Bronze Age scholarship and to the British Museum’s current strategy document: Towards 2020: The Museum for the Global Citizen (BM ‘Towards 2020’).5 Methodological details and information on initial results are also given, as they may prove useful to those undertaking similar archaeological archive projects within museums. 1 http://archaeologydataservice.ac.uk, accessed 28/2/15 http://www.data-archive.ac.uk, accessed 28/2/15 3 http://opencontext.org, accessed 28/2/15 4 http://www.tdar.org, accessed 28/2/15 5 https://www.britishmuseum.org/pdf/Towards_2020-The_British_Museum_Strategy.pdf, accessed 28/2/15 2 The MicroPasts Project and the British Bronze Age One of the MicroPasts projects key aims, to create new open access datasets through crowd-sourcing, is particularly relevant to the study of Bronze Age material culture. Artefact studies were once the cornerstone of the study of British later prehistory, particularly during the 1960s to 1980s, when students of key figures (in what we can lump together as a Culture Historical stage in Western world archaeology) worked to produce important studies of ceramics and metalwork. Examples of these outputs include key volumes on metalwork and ceramics appearing in Prähistorische Bronzefunde and the Gulbenkian Archaeological Series, and early volumes of British Archaeological Reports (see Brindley 2008). As post-processual archaeologies grew in stature and influence from the mid-1980s onwards, the importance of artefact studies waned in Britain and was marked by a growing wariness of the traditionalism of artefact and (particularly) typological studies (cf. Brück 2008, 28-9). From this time until recently, when a return to artefact-led studies seems manifest (Hicks 2010; Joy 2010; Hodder 2012), new discoveries have continued to be made, particularly as the result of the Portable Antiquities Scheme (PAS)6 and changes to the UK Treasure legislation (The Treasure Act 1996, and most notably the implications of the designation order of 2002: see Bland 2005; Murgia et al. 2015). Yet there has been little attempt to digest the implications of this deluge of new discoveries for our current understanding of the period (Roberts 2008, 531-3), or to integrate old datasets with new discoveries (cf. Murgia et al. 2015). The years of relative inactivity in dataset creation for British Bronze Age artefacts together with the rapid development of online resources present a myriad of opportunities. An important example is the digitising, transcription and enhancement of the NBAI. A brief biography of the National Bronze Age Implement Index The NBAI developed primarily from the 1910s to the 1980s, when the resource gradually stopped being updated and maintained. From that point of arrested development until now, it has consisted of approximately 30,000 double sided-cards covering Bronze Age ‘implements’ (primarily weapons, tools and ornaments) arranged by object type in drawers and then county in several BM filing cabinets (Fig. 1). The original impetus came from the British Association for the Advancement of Science (BAAS), who established a Bronze Implement Committee to compile a catalogue of Bronze Implements (Myres 1920). They argued that understanding the distribution patterns of Bronze Age implements could solve many of the research questions faced by prehistorians trying to understand the period and they proposed the construction of a complete catalogue of ‘all the metal objects of the Bronze Age in the museums and collections in the British Isles’ (Ibid.). The work, coordinate by the secretary of the Committee, Harold Peake, proved to be a collective inter-museum and public effort as it involved writing to museums, collections and individuals who might hold relevant information. It was partly funded by the BAAS, but Myres (1920) records that ‘[g]enerous donors’ had ‘subscribed more than £50 in the year 1919-20’, and called for: ‘public attention to the valuable work already accomplished, and to the need of further 6 https://finds.org.uk/, accessed 28/2/15 help in order to complete it’. The crowd-sourcing of the Index by the MicroPasts project is thus only the latest stage in the collaboration of professionals and the wider public that stretches back for almost a century. [Figure 1: Image of the National Bronze Age Implement Index] By 1933, work on the NBAI was, by the BAAS’s reckoning, almost complete, and they offered the Index to the Trustees of the British Museum on condition that it was kept up to date and made accessible to students, which they duly accepted (Antiq J 1934, 178). At the BM the Index was kept within the department of British and Mediaeval Antiquities, where the noted prehistorian, Christopher Hawkes was an active Assistant Keeper between 1928 and 1946. When Hawkes moved to Oxford to become the first Professor of European Archaeology in 1946, the Index followed him on loan (British Museum NBAI history file). The Index served as an important research tool to enable Hawkes, his research assistant, Margaret Smith, and students (including Denis Britton), to produce a number of seminal studies and catalogues of Bronze Age metalwork, including contributions to the important, European, Inventaria Archaeologica series (see Burgess 2004, 347-8; Díaz-Andreu et al. 2009, 410-11). This period of the Index’s history marks a shift in its accessibility and function: from a public tool to one of more personal and academic study. In 1966, the loan of the NBAI ended and it was returned to the British Museum. The BM archives record that at the creation of the Department of Prehistoric and Romano-British Antiquities in 1969, ‘there was still a build-up of staff to take place before all the services we want (such as the efficient upkeep of the national index of Bronze Implements) can be carried out’ (British Museum NBAI history file). New work on the Index started in 1973 and continued throughout that decade, before gradually declining over the course of the 1980s. In the 1980s, it was recognised by Stuart Needham that the NBAI was an important resource but that its format and location made it unfriendly to users out with the BM (Ibid.). By the 1990s, Needham had also recognised that some form computerisation was ‘long favoured’, but noted that computerisation would not serve to make the database any more accessible than the NBAI itself (Ibid.). In hindsight, the task of making the Index as an open and searchable resource required the development of the participatory Internet, of high-speed scanners and a crowd-sourced platform. The Index remained at the BM’s Orsman Road store until it was moved to the BM’s main Bloomsbury site for digitising as part of the MicroPasts project in early 2014. The process of digitising and transcribing the Index via MicroPasts during 2014 and 2015 has opened it to a wider audience and hopefully the legacy of a digitised and transcribed dataset, available online, will endure for even longer. Methodology The Index cards have been scanned using a high capacity, A3 scanner from Canon, to produce images at 600 dpi. The front and back of the cards are then merged into a single file via a simple Python script by our team (DP & AB), after which the images are uploaded to the project’s ‘Flickr’ webpage7, Amazon S3 storage and the BM’s PAS server farm (backup in triplicate). Individual applications (‘apps’) are constructed by 7 https://www.flickr.com/photos/micropasts, accessed 28/2/15 project team members, typically based on the theme of the NBAI’s many drawers, for instance ‘Drawer A2 – Socketed Spearheads’. This helps to make the tasks more manageable and from a public participant’s perspective, more interesting and bite-sized, with the ability to see tasks being achieved thanks to their assistance, rather than being faced with a long, characterless and seemingly never-ending progress bar. The transcription interface prompts users to fill in fields based on the contents of the cards, and once the app is complete, a script (using ‘R’) is run to consolidate the answers into a single csv file; and members of the project team examine the results for errors and to confirm the final transcription. The data is always available to the wider public through the project website, regardless of what stage it is in that process: the scanned cards on the ‘Flickr’ website and the transcribed data on the project website. Social media (Twitter and Facebook) have also played a role in spreading raising awareness of currently available apps. The decision was made at the outset of the project that public contributors need to be able to see and make free use of the work that they have achieved, so they and other MicroPasts outputs are licensed under Creative Commons CC08 or CC-BY9. Initial results and wider implications At the time of writing (late February 2015) around 17,000 cards (double sided) have been scanned and transcribed by around 1000 contributors. While some contributors provide transcriptions of a few cards, others do many. All projects of this type involve a degree of the risk and uncertainty, but the pace of public participation has surprised us. A study and analysis of patterns of participation is underway (Bonacchi 2015; Bonachhi et al. forthcoming) and is therefore not rehearsed here. The process of digitising and transcribing the NBAI has provided us with a range of information. Each card contains information relating to: • • • • • • Place name and spatial information; The circumstances of their discovery; Present whereabouts (personal and museum collection); Bibliographical information; Written descriptions (including measurements); Context of discovery (with later cards and those embellished at a later date having more in the way of information); On the reverse sides of many of the cards are line illustrations (usually measured drawing) of the object in question (Fig. 2). The NBAI data can be compared to, and serve to complement, the more recently discovered Bronze Age metal objects recorded in the PAS database (by a generally non-specialist recording agent.) As the respective datasets were compiled under different conditions and at different times, some factors must be heeded. The NBAI was mainly formed in the 18th and 19th centuries through casual discoveries in agricultural operations, whilst the PAS database has been built over the last two decades, primarily 8 9 http://creativecommons.org/publicdomain/zero/1.0/, accessed 28/2/15 http://creativecommons.org/licenses/by/4.0/, accessed 28/2/15 via discoveries made by metal detectorists. Detailed comparison may serve to highlight important biases in patterns of collection and survival and to validate other patterns as real reflections of Bronze Age activity. The prospect of combining datasets raises the possibility of working towards a ‘total’ dataset of metalwork finds that is open and searchable. This aspiration is far from new, but the feasibility of exhaustive recording is now much higher, we would argue, and would enhance our ability to identify some of the fundamental patterns and trends in metalwork deposition, the type of social and cultural insights that the NBAI was originally established to address. It is arguably only now, with the assistance of the crowd and the availability of the Internet that this can be achieved. There is a pleasing symmetry to this process. A project that started by calling for the participation of the many, but survived fitfully on the efforts of a few, has come full circle and is now being made digitally available for all thanks to the advent of new technology, methodology and many willing and able volunteer transcribers: ‘citizen archaeologists’ (Doherty 2014).10 [Figure 2: An example of a card from the Index] 3D Models of Bronze Age objects An additional component of the project, indirectly related to the NBAI, has been the coproduction of 3D models of objects of Bronze Age in the British Museum collection. Thanks to considerable improvements in 3D modelling techniques, it is now relatively straightforward to produce high quality 3D models without the use of expensive equipment or software, using only digital photographs taken at various points around an object in a technique known as structure-from-motion (SfM). However, for best results, the objects under study need to be ‘masked’, to pick out the object in question from from its background. This is achieved by MicroPasts contributors who draw around each object depicted in a photo within the crowd-sourcing platform. There after, the production of the finished model is undertaken by members of the project team and a number of trained/experienced volunteers or ‘mediators’. The models are then hosted on the Sketchfab website (and within the MicroPast project’s own WebGL environment), designed to display and share 3D content online. So far, nearly 60 models of Bronze Age British Museum objects have been produced11, with many of the final models having been produced by a ‘citizen archaeologist’. Sketchfab also allows for models to be annotated and thus enables users to take ‘tours’ of the objects, zooming and moving between particular camera positions of interest, discursively-enhanced landmarks, an overall title and a description (up to a maximum of 256 characters) (Fig. 3). The annotation function has considerable potential for how objects can be investigated and explored within existing online collection databases. Instead of online visitors being faced with photographs of museum objects accompanied by separate pages of text, the annotations combine image and description in a more integrated way and are more like the experience of a hands-on tour or handling tour, transforming untargeted text into a curator’s eye view of the objects. The annotated models could thus serve to enhance the experience of existing permanent displays by conveying both a volume of information, detail and multiple angles that are not possible 10 11 A term also developed by SCAPE: http://www.scharp.co.uk/, accessed 28/2/15 https://sketchfab.com/micropasts/folders, accessed 28/2/15 within galleries or through the medium of traditional museum labels. The same model could be given multiple tours aimed at different age or interest groups. From February 2015, the PAS website embeds models within its records (for example: GLO-E9EC1612) via the Sketchfab oEmbed13 application programming interface (API). The inclusion of these models provides a richer experience; comprehensive descriptive text, high quality images (and with zoom), maps and 3D downloadable models all made available under a liberal CC-BY license. This may be a model for museums collections databases to pursue, to enable new types of analytical scholarship. [Figure 3: Using Sketchfab to annotate SfM models generated by the MicroPasts project] The British Museum and the MicroPasts Project The collaboration between the British Museum and MicroPasts is relevant to several aspects of the BM’s current strategy, as outlined in ‘Towards 2020’. The original Age of Enlightenment vision of the Museum was of a collection ‘available, free of charge, to all visitors, native and foreign’ (BM Towards 2020, 2; see Sloan & Burnett (eds) 2003), and the ‘Towards 2020’ document highlights the success of Collection Online14 in making over two million objects available worldwide while expressing a desire to move forward with the process of making collections available online in new ways. Indeed, the British Museum is visited online far more often than in person: 24 million visitors compared with approximately 6 million visitors in person in 2011-12 (BM Towards 2020, 4). The MicroPasts project has sought to do so by digitising versions of records that concern discoveries of Bronze Age objects from a range of other museums and private collections, not just those within the BM itself. Together with the PAS database, the project is contributing to one of the largest and most integrated databases of Bronze Age implements in the world. Furthermore, the use of public, crowd-sourced assistance in the production of SfM 3D models serves as a test of concept for how museums can enhance the online presentation of objects to the public at an economically challenging time and in a way that engages the public in the actual process of database construction. Final thoughts It is hoped that the scanning of the NBAI will be completed, and that the vast majority of transcription will have been completed by the end of the project (in April 2015). The next steps concern reviewing, consolidating, updating and integrating the NBAI with the PAS database and with the BM database, and future projects are planned along these lines. The process has served as a test of concept for the methodology of crowd-sourcing this type of data but has also proved valuable from a curatorial point of view, in terms of finding and maintaining an audience or re-connecting with an existing one in new ways. At the outset of this paper we highlighted the link that Ascherson (2002) drew between national identities and the degree or order and organisation applied to accounts and 12 https://finds.org.uk/database/artefacts/record/id/627280, accessed 28/2/15 https://sketchfab.com/developers/oembed, accessed 28/2/15 14 http://www.britishmuseum.org/research/collection_online/search.aspx, accessed 28/2/15 13 material relating to the past. One of the most exciting elements of open, crowd-sourced datasets is that it provides space and freedom for flexibility of order and organisation. We can have the best of both worlds without imposing one or the other on the public whose interests lie in the material we happen to study professionally. Individual cards can serve as discoveries with personal or anecdotal significance or have particular local/regional significance but can also be analysed and interpreted by professional archaeologists working on broader themes. Furthermore, the NBAI reflects some of the major changes in the character and trajectory of the British Bronze Age studies and academic and museum archaeology more generally (cf. Wexler et al. forthcoming). There is clearly much still to learn from our archives and from the legacy of the museum archaeologists who go before us. Bibliography Antiq J 1934. ‘Notes’, Antiquaries Journal, 14, 178 Ascherson, N. 2002. Stone Voices. The Search for Scotland. London: Granta Books Bevan, A. 2012. Value, authority and the open society. Some implications for digital and online archaeology, in C. Bonacchi (ed.) Archaeology and Digital Communication: Towards Strategies of Public Engagement, 1-14, London: Archetype Bevan, A., C. Bonacchi, D. Pett, A. Keinan-Schoonbaert, J. Wexler & N. Wilkin, 2014. ‘Citizen archaeologists. Online collaborative research about the human past’. Human Computation 1.2: 183-197 Bland, R. 2005. A pragmatic approach to the problem of portable antiquities: the experience of England and Wales’, Antiquity, 79, 440-7 Bonacchi, C., A. Bevan, D. Pett, A. Keinan-Schoonbaert, R. Sparks, J. Wexler & N. Wilkin, 2014. Crowd-sourced Archaeological Research: The MicroPasts Project. Archaeology International, No. 17, 61-68 Bonacchi, C. 2015. Engaging the Public: The Impact of a Public Archaeology. Seminar presentation. Research Seminar Series of the Department of Archaeology, Reading University, 23 February 2015 [online]. Available at: https://www.academia.edu/11040122/Engaging_the_Public_the_impact_of_a_Public_Ar chaeology (Accessed 28/2/15) Bonacchi, C., Bevan, A., Pett, D., Keinan-Schoonbaert, A. Forthcoming (submitted). Early experiments in crowd-funding community archaeology. Journal of Community Archaeology and Heritage. Brindley, A.L. 2008. A Canon for the Bronze Age? Bronze Age Review, 1, 1–6 (=BM ‘Towards 2020) The British Museum. Towards 2020: The British Museum’s Strategy, https://www.britishmuseum.org/pdf/Towards_2020The_British_Museum_Strategy.pdf (Accessed 28/2/15) Brück, J. 2008. Prospects and potential in the archaeology of Bronze Age Britain. Bronze Age Review, 1, 23-33 Burgess, C. 2004. ‘Forty-five years’ researches in the chronology and ordering of the British Bronze Age: a personal memoir. In A. Gibson & A. Sheridan (eds), From Sickles to Circles: Britain and Ireland at the time of Stonehenge, Stroud: Tempus, 339-65 Cooper, A. & Green, C. forthcoming. Embracing the complexities of archaeological ‘big data’: the case of the English Landscape and Identities project. Journal of Archaeological Method and Theory Díaz-Andreu, M. Price, M. & Gosden, C. 2009.Christopher Hawkes: His Archive and Networks in British and European Archaeology. The Antiquaries Journal, 89, 405-26 Doherty, J. 2014. Crowdsourcing the Bronze Age, http://pybossa.com/blog/2014/06/05/crowdsourcing-the-bronze-age/ (Accessed 28/2/15) Hicks, D. 2010. The Material-Cultural Turn: Event and Effect. In D. Hicks and M. C. Beaudry (eds), The Oxford Handbook of Material Culture Studies. Oxford: Oxford University Press, 25- 98 Hole, B. 2012. ‘A Call for Open Scholarship in Archaeology’, in C. Bonacchi (ed.) Archaeology and Digital Communication: Towards Strategies of Public Engagement, 114-126, London: Archetype Hodder, I. 2012. Entangled: An Archaeology of the Relationships Between Humans and Things. Malden, MA: Wiley-Blackwell Joy, J. 2010. Reflections on the Iron Age: biographies of mirrors. British Archaeological Reports (British Series) 518, Oxford: Archaeopress Murgia, A., Roberts, B. & Wiseman, R. 2015. What Have Metal-Detectorists Ever Done for Us? Discovering Bronze Age Gold in England and Wales. Archäologisches Korrespondenzblatt, 44, 353-67 Myres, J.L. 1920, A Catalogue of Bronze Implements’, The Spectator, 18 December, p.12. http://archive.spectator.co.uk/article/18th-december-1920/12/a-catalogue-ofbronze-implements (Accessed 28/2/15) Newman, M. 2011. The database as material culture. Internet Archaeology, 29. http://intarch.ac.uk/journal/issue29/tag_index.html (Accessed 28/2/15) Onsrud, H. & Campbell, J. 2007. Big opportunities in access to “small science” data. Data Science Journal, 6 (Open Data Issue), 58-66 Roberts, B. 2008. Information, interaction and society. Bronze Age Review, 1, 48-56 Sloan, K. & Burnett, A. 2003. Enlightenment : discovering the world in the eighteenth century, London: The British Museum Press Wexler, J., Bevan, A., Bonacchi, C. Keinan-Schoonbaert, A., Pett, D. & Wilkin, N. Forthcoming. Collective Re-Excavation and Lost Media from the Last Century of British Prehistoric Studies, Journal of Contemporary Archaeology (Forum on ‘Media Archaeologies’) Fig. 1 Fig. 2 Fig. 3